CSE 486/586 CSE 486/586 Distributed Systems Leader Election Steve Ko Computer Sciences and...
-
Upload
brandon-stanley -
Category
Documents
-
view
218 -
download
0
description
Transcript of CSE 486/586 CSE 486/586 Distributed Systems Leader Election Steve Ko Computer Sciences and...
CSE 486/586
CSE 486/586 Distributed SystemsLeader Election
Steve KoComputer Sciences and Engineering
University at Buffalo
CSE 486/586 3
Why Election?• Example 1: sequencer for TO multicast• Example 2: leader for mutual exclusion• Example 3: group of NTP servers: who is the root
server?
CSE 486/586 4
What is Election?• In a group of processes, elect a leader to undertake
special tasks. • What happens when a leader fails (crashes)
– Some process detects this (how?)– Then what?
• Focus of this lecture: election algorithms – 1. Elect one leader only among the non-faulty processes– 2. All non-faulty processes agree on who is the leader
• We’ll look at 3 algorithms
CSE 486/586 5
Assumptions• Any process can call for an election.• A process can call for at most one election at a time.• Multiple processes can call an election
simultaneously.– All of them together must yield a single leader only– The result of an election should not depend on which
process calls for it.• Messages are eventually delivered.
CSE 486/586 6
Problem Specification• At the end of the election protocol, the non-faulty
process with the best (highest) election attribute value is elected. – Attribute examples: CPU speed, load, disk space, ID– Must be unique
• Each process has a variable elected.• A run (execution) of the election algorithm must
always guarantee at the end:– Safety: non-faulty p: (p's elected = (q: a particular non-
faulty process with the best attribute value) or )– Liveness: election: (election terminates) & p: non-faulty
process, p’s elected is eventually not
CSE 486/586 7
Algorithm 1: Ring Election[Chang & Roberts’79] • N Processes are organized in a logical ring
– pi has a communication channel to pi+1 mod N.– All messages are sent clockwise around the ring.
• To start election– Send election message with my ID
• When receiving message (election, id)– If id > my ID: forward message
» Set state to participating– If id < my ID: send (election, my ID)
» Skip if already participating» Set state to participating
– If id = my ID: I am elected (why?) send elected message» elected message forwarded until it reaches leader
CSE 486/586 8
Ring-Based Election: Example• The worst-case
scenario occurs when?– the counter-clockwise
neighbor (@ the initiator) has the highest attr.
• In the example: – The election was
started by process 17.– The highest process
identifier encountered so far is 24
– (final leader will be 33)
24
15
4
33
28
17
24
1
9
CSE 486/586 9
Ring-Based Election: Analysis• In a ring of N processes,
in the worst case:– N-1 election messages to
reach the new coordinator
– Another N election messages before coordinator decides it’s elected
– Another N elected messages to announce winner
• Total Message Complexity = 3N-1
• Turnaround time = 3N-1
24
15
9
4
33
28
17
24
1
CSE 486/586 10
Correctness?• Safety: highest process elected• Liveness: complete after 3N-1 messages
– What if there are failures during the election run?
CSE 486/586 11
Example: Ring Election
Election: 2
Election: 4
Election: 4 Election: 3
Election: 4
P1
P2
P3
P4
P0
P5
1. P2 initiates election after old leader P5 failed
P1
P2
P3P4
P0
P5
2. P2 receives "election", P4 dies
P1
P2
P3P4
P0
P5
3. Election: 4 is forwarded forever?
May not terminate when process failure occurs during the election!
Consider above example where attr==highest id
CSE 486/586
CSE 486/586 Administrivia• PA2 due next week• Midterm: 3/11 (Wednesday) in class
– Multiple choices– Everything up to today
12
CSE 486/586 13
Algorithm 2: Modified Ring Election • election message tracks all IDs of nodes that
forwarded it, not just the highest– Each node appends its ID to the list
• Once message goes all the way around a circle, new coordinator message is sent out– Coordinator chosen by highest ID in election message– Each node appends its own ID to coordinator message
• When coordinator message returns to initiator– Election a success if coordinator among ID list– Otherwise, start election anew
CSE 486/586
Example: Ring Election
Election: 2
Election: 2, 3,4,0,1
Election: 2,3,4Election: 2,3
Coord(4): 2
Coord(4): 2,3
Coord(4) 2, 3,0,1
Election: 2
Election: 2,3
Election: 2,3,0
Election: 2, 3,0,1
Coord(3): 2
Coord(3): 2,3
Coord(3): 2,3,0
Coord(3): 2, 3,0,1
P1
P2
P3
P4
P0
P5
1. P2 initiates election
P1
P2
P3P4
P0
P5
2. P2 receives "election", P4 dies
P1
P2
P3P4
P0
P5
3. P2 selects 4 and announces the result
P1
P2
P3P4
P0
P5
4. P2 receives "Coord", but P4 is not included
P1
P2
P3P4
P0
P5
5. P2 re-initiates election
P1
P2
P3P4
P0
P5
6. P3 is finally elected
14
CSE 486/586 15
Modified Ring Election• How many messages?
– 2N• Is this better than original ring protocol?
– Messages are larger• Reconfiguration of ring upon failures
– Can be done if all processes "know" about all other processes in the system
• What if initiator fails?– Successor notices a message that went all the way around
(how?)– Starts new election
• What if two people initiate at once– Discard initiators with lower IDs
CSE 486/586 16
What about that Impossibility?• Can we have a totally correct election algorithm in a
fully asynchronous system (no bounds)– No! Election can solve consensus
• Where might you run into problems with the modified ring algorithm?– Detect leader failures– Ring reorganization
CSE 486/586 17
Algorithm 3: Bully Algorithm • Assumptions:
– Synchronous system– attr=id– Each process knows all the other processes in the system
(and thus their id's)
CSE 486/586 18
Algorithm 3: Bully Algorithm • 3 message types
– election – starts an election– answer – acknowledges a message– coordinator – declares a winner
• Start an election– Send election messages only to processes with higher IDs
than self– If no one replies after timeout: declare self winner– If someone replies, wait for coordinator message
» Restart election after timeout• When receiving election message
– Send answer– Start an election yourself
» If not already running
CSE 486/586 19
Example: Bully Election
OKOK
P1
P2
P3
P4
P0
P5
1. P2 initiates election 2. P2 receives replies
P1
P2
P3
P4
P0
P5
3. P3 & P4 initiate election
P1
P2
P3
P4
P0
P5
P1
P2
P3
P4
P0
P5
4. P3 receives reply
OK
ElectionElection
Election
ElectionElection
Election
P1
P2
P3
P4
P0
P5
5. P4 receives no reply
P1
P2
P3
P4
P0
P5
5. P4 announces itself
coordinator
answer=OK
CSE 486/586 20
The Bully Algorithm
The coordinator p4 fails and p1 detects this
p3 fails
p1 p2
p3
p4
p1
p2
p3
p4
Ccoordinator
Stage 4
C
election
electionStage 2
p1
p2
p3
p4
C
election
answer
answer
electionStage 1
timeout
Stage 3
Eventually.....
p1
p2
p3
p4
election
answer
election
CSE 486/586 21
Analysis of The Bully Algorithm• Best case scenario?• The process with the second highest id notices the
failure of the coordinator and elects itself.– N-2 coordinator messages are sent.– Turnaround time is one message transmission time.
CSE 486/586 22
Analysis of The Bully Algorithm• Worst case scenario?• When the process with the lowest id in the system
detects the failure.– N-1 processes altogether begin elections, each sending
messages to processes with higher ids.– The message overhead is O(N2).
CSE 486/586 23
Turnaround time• All messages arrive within T units of time
(synchronous)• Turnaround time:
– election message from lowest process (T)– Timeout at 2nd highest process (X)– coordinator message from 2nd highest process (T)
• How long should the timeout be?– X = 2T + Tprocess
– Total turnaround time: 4T + 3Tprocess
CSE 486/586 24
Summary• Coordination in distributed systems sometimes
requires a leader process• Leader process might fail• Need to (re-) elect leader process• Three Algorithms
– Ring algorithm– Modified Ring algorithm– Bully Algorithm